JP6418961B2 - Obstacle detection device, moving object, obstacle detection method, and obstacle detection program - Google Patents

Description

  The present invention relates to an obstacle detection device and a moving body that include a distance measuring sensor that measures a distance to an object to be detected, and an obstacle detection method and an obstacle detection program thereof.

  Conventionally, in a system that autonomously travels on a traveling surface, an obstacle is detected by a sensor or the like to control the operation. For example, in a vehicle such as an automobile, a vehicle contact avoidance support device that supports contact avoidance with an obstacle has been developed as a system for assisting a driver (see, for example, Patent Document 1).

  In mobile robots, a method for preventing contact with an obstacle by moving in a direction instructed by remote operation and calculating a moving speed based on a distance from the obstacle detected by a sensor has been proposed. (For example, refer to Patent Document 2).

JP 2008-49932 A JP 2006-285548 A

  The vehicle contact avoidance assistance device described in Patent Document 1 detects a distance to an obstacle ahead of the vehicle by a front obstacle sensor, calculates a target speed of the vehicle according to the distance to the obstacle, The braking force is controlled based on the speed, and the vehicle speed is reduced by automatic braking. Therefore, in the contact avoidance support device, the target speed is corrected according to the operation state of the vehicle by the driver, and the driver avoids a sense of incongruity by assisting the driver's spontaneous contact avoidance operation. To avoid contact with

  In addition, the mobile robot described in Patent Document 2 includes a detection unit that detects an obstacle ahead of the movement direction, calculates the distance to the obstacle, and is instructed by the monitoring center about the direction and speed of movement. The When the mobile robot detects a front obstacle in the autonomous movement mode, the mobile robot transmits an abnormal signal to the monitoring center. In the monitoring center, remote control of the mobile robot is performed by the controller. Then, in the remote operation mode, the mobile robot moves at a traveling speed that can be stopped before it touches the obstacle in accordance with an instruction from the monitoring center. As described above, in the mobile robot using the remote operation, the moving speed is set in consideration of obstacle avoidance so that the instruction delay due to the transmission delay can be dealt with.

  By the way, a system using autonomous traveling such as a mobile robot or a vehicle is used not only indoors but also outdoors, and is required to cope with various environments. In the outdoors, changes in the weather are taken into account, and there is a problem that raindrops and the like at the time of rainfall are erroneously detected as obstacles. Patent Document 1 and Patent Document 2 described above do not consider misdetecting an obstacle due to raindrops or the like, and there is a problem that it is not possible to distinguish between an obstacle that should avoid contact and a raindrop that does not prevent movement. is there. In addition, a process for removing noise from the detection result has been proposed so that raindrops and the like are not erroneously detected as an obstacle, but if such a process is performed during normal times, a small obstacle may be overlooked. .

  The present invention has been made in order to solve the above-described problem, and during rainfall, an obstacle detection device, a moving body, an obstacle detection method capable of avoiding erroneous detection of raindrops as an obstacle, And to provide an obstacle detection program.

  An obstacle detection device according to the present invention includes a distance measuring sensor that transmits a detection wave to an object to be detected, receives a reflected wave from the object to be detected, and measures a detection distance to the object to be detected; A detection image generation unit that generates a detection image indicating the presence or absence of the detection object in the detection range in which the detection wave is transmitted based on a distance measurement result of the distance measurement sensor; and the object to be detected in the detection image. A rain determining unit that determines whether or not the detection range is in a rain state based on the presence or absence of discrete points that are discretely located among the detected objects, and a rain that removes the isolated points from the detected image A rain removal processing unit that performs a removal process, and an obstacle determination unit that determines whether the detected object is an obstacle based on the detected image, and the obstacle determination unit includes: If it is determined by the rain determination unit that the rain state is present, the rain It based the processed section into the rain removing processed sensed image, which comprises carrying out the obstacle determining.

  In the obstacle detection device according to the present invention, the detection range is a three-dimensional space, and the detection image generation unit generates a three-dimensional image corresponding to the detection range that is a three-dimensional space as the detection image. The rain determination unit may perform the rain determination on the three-dimensional image.

  In the obstacle detection device according to the present invention, the rain determination unit may calculate a variance value of the detected object in the detection image at the time of the rain determination.

  In the obstacle detection device according to the present invention, the rain determination unit may calculate a difference in the detection distance from the periphery of the detection object in the detection image at the time of the rain determination.

  In the obstacle detection device according to the present invention, the detection image generation unit may be configured to periodically generate the detection image for each preset generation cycle.

  In the obstacle detection device according to the present invention, the rain removal processing unit may perform the rain removal processing using a median filter.

  In the obstacle detection device according to the present invention, the rain determination unit compares the detection image with a median-processed image obtained by removing the isolated points using a median filter with respect to the detection image, and performs the rain determination. It is good also as a structure to perform.

  In the obstacle detection device according to the present invention, the rain determination unit may perform the rain determination by comparing two detection images generated in different generation periods.

  In the obstacle detection device according to the present invention, the rain determination unit may perform the rain determination for each of the plurality of generation periods.

  In the obstacle detection device according to the present invention, the rain determination unit performs the rain determination at a plurality of the generation periods, and outputs the result of the rain determination when the same determination is made a plurality of times in succession. Also good.

  In the obstacle detection device according to the present invention, the rain removal processing unit may be configured to remove the isolated points by comparing the detection distances for each of the plurality of generation periods with respect to the detection image.

  A moving body according to the present invention includes the obstacle detection device according to the present invention and travels on a traveling surface.

  The mobile body according to the present invention includes a travel control unit that controls a travel speed that travels on the travel surface, and the travel control unit, when the rain determination unit determines that the rain state is present, the travel speed It is good also as a structure which controls to decelerate.

  The obstacle detection method according to the present invention includes a distance measuring sensor that transmits a detection wave to an object to be detected, receives a reflected wave from the object to be detected, and measures a detection distance to the object to be detected. An obstacle detection method in an obstacle detection apparatus provided, wherein the presence or absence of the detection object in a detection range in which the detection wave is transmitted to a detection image generation unit based on a distance measurement result of the distance measurement sensor The detection range is generated in a rainy state based on the presence or absence of isolated points discretely located among the detected objects in the detection image. A rain determining step for determining whether there is a rain, a rain removing processing step for causing a rain removing processing unit to perform a rain removing process for removing the isolated point from the detected image, and an obstacle judging unit for detecting the detection Based on the image, the detected object An obstacle determination step for making an obstacle determination as to whether or not the object is an obstacle, and when the obstacle determination step is determined to be in the rain state in the rain determination step, the rain removal processing step includes the step The obstacle determination is performed based on the detection image subjected to the rain removal process.

  The obstacle detection program according to the present invention causes a computer to execute each step of the obstacle detection method according to the present invention.

  According to the present invention, it is possible to obtain a detection image showing only an obstacle by performing rain removal processing on the detection image during the rain, and it is possible to avoid erroneous detection of raindrops as an obstacle.

It is an external view of the moving body which concerns on 1st Embodiment of this invention. It is a block diagram of the moving body shown in FIG. It is explanatory drawing which shows the 1st detection image produced | generated by the detection image generation part. It is explanatory drawing which shows the 2nd detection image produced | generated by the detection image generation part. It is explanatory drawing which shows the 3rd detection image produced | generated by the detection image generation part. It is explanatory drawing which shows the 4th detection image by which the rain removal process part performed the rain removal process. It is explanatory drawing which shows the 5th detection image by which the rain removal process part performed the rain removal process. It is a flowchart which shows the processing flow of the obstacle detection method of the obstacle detection apparatus which concerns on 1st Embodiment of this invention. It is a flowchart which shows the processing flow of the rain determination process 1 in 1st Embodiment of this invention. It is a flowchart which shows the processing flow of the rain removal process 1 in 1st Embodiment of this invention. It is explanatory drawing which shows the 6th detection image produced | generated by the detection image generation part in 2nd Embodiment of this invention. It is explanatory drawing which shows the 7th detection image produced | generated by the detection image generation part in 3rd Embodiment of this invention. It is explanatory drawing which shows the difference image which compared the 1st detection image and the 7th detection image. It is a flowchart which shows the processing flow of the rain determination process 2 in 3rd Embodiment of this invention. It is explanatory drawing which shows the relationship between the production | generation period and rainfall determination period in 4th Embodiment of this invention. It is a flowchart which shows the processing flow of the obstacle detection method of the obstacle detection apparatus which concerns on 4th Embodiment of this invention. It is explanatory drawing which shows the relationship between the repetition of the rain determination in 5th Embodiment of this invention, and the output of a determination result. It is a flowchart which shows the processing flow of the obstacle detection method of the obstacle detection apparatus which concerns on 5th Embodiment of this invention. It is a flowchart which shows the processing flow of the continuous determination process in 5th Embodiment of this invention. It is explanatory drawing of the 1st distance data which shows a detection distance. It is explanatory drawing of the 2nd distance data which shows a detection distance. It is explanatory drawing which shows the 9th detection image produced | generated by the detection image generation part. It is explanatory drawing which shows the 10th detection image produced | generated by the detection image generation part. It is a flowchart which shows the processing flow of the rain determination process 3 in 3rd Embodiment of this invention. It is explanatory drawing which shows the 3rd distance data in a 1st frame. It is explanatory drawing which shows the 4th distance data in a 2nd frame. It is explanatory drawing which shows the 5th distance data in a 3rd frame. It is explanatory drawing which shows the 6th distance data which acquired the maximum value of the detection distance. It is a flowchart which shows the process flow of the rain removal process 2 in 7th Embodiment of this invention.

(First embodiment)
Hereinafter, an obstacle detection device and a moving body according to a first embodiment of the present invention will be described with reference to the drawings.

  FIG. 1 is an external view of a moving body according to the first embodiment of the present invention, and FIG. 2 is a configuration diagram of the moving body shown in FIG.

  The moving body 1 according to the first embodiment of the present invention transmits a detection wave to the detected object 50, receives a reflected wave from the detected object 50, and sets a detection distance KL to the detected object 50. The vehicle includes an obstacle detection device 10 provided with a distance measuring sensor 11 to be measured, and travels on a traveling surface 100. Specifically, the moving body 1 is a four-wheeled vehicle that moves along a preset route, and includes a distance measuring sensor 11, a drive unit 20, and a CPU 30.

  The drive unit 20 includes four wheels and a drive source such as a motor. The drive unit 20 is not limited to this, and the number of wheels may be changed, a belt or the like may be used, and the driving unit 20 may be configured to travel and adjust the traveling speed as appropriate. That's fine. The drive unit 20 is controlled in speed and direction of travel by the travel control unit 30e.

  The distance measuring sensor 11 is an optical sensor using LIDAR (Light Detection and Ranging), irradiates a laser beam as a detection wave over a wide range, and receives reflected light (reflected wave) from an object. Thereby, the distance measuring sensor 11 detects the position of the detected object 50 and the distance to the detected object 50. The detection result of the distance measuring sensor 11 will be described in detail with reference to FIGS. 3A to 3C described later.

  The CPU 30 stores the detected image generation unit 30a, the rain determination unit 30b, the rain removal processing unit 30c, the obstacle determination unit 30d, and the travel control unit 30e as preinstalled programs, and executes the stored program. Thus, the processing described later is executed. The CPU 30 is used in common by the moving body 1 and the obstacle detection device 10. Although omitted in FIG. 2, the moving body 1 may be provided with a storage device for storing the detection result of the distance measuring sensor 11 and various settings such as the traveling speed.

  FIG. 1 shows a moving body 1 that is traveling in a raining state where it is raining. The distance measuring sensor 11 has a detection range KH set along the traveling direction of the moving body 1, and detects the raindrop 52 as the detected object 50 in addition to the obstacle 51 in the detection range KH.

  FIG. 3A is an explanatory diagram illustrating a first detection image generated by the detection image generation unit, and FIG. 3B is an explanatory diagram illustrating a second detection image generated by the detection image generation unit, and FIG. These are explanatory drawings which show the 3rd detection image produced | generated by the detection image generation part.

  The distance measuring sensor 11 is configured to output the position of the detected object 50 in the detection range KH, and the detection image generating unit 30a is detected in the detection range KH based on the distance measurement result of the distance measuring sensor 11. A detection image (for example, first detection image GZ1) indicating the presence or absence of the object 50 is generated. The detected image is an image in which the presence / absence of the detected object 50 in the detection range KH is associated with the position of the detected object 50. In the present embodiment, the detected image shows a plane orthogonal to the traveling direction of the moving body 1 and has a rectangular shape. In the detected image, the detected object 50 is arranged as a reflection point at coordinates (pixels) corresponding to the position, the pixel (reflective point) where the detected object 50 exists is “1”, and the pixel without the detected object 50 is “ 0 ”, which is regarded as binarized image data. In the detection images shown below (first detection image GZ1 to tenth detection image GZ10), the horizontal direction is indicated as the horizontal direction X, and the direction perpendicular to the horizontal direction X is indicated as the vertical direction Y. In the following, for simplification of description, a pixel without the object 50 (a pixel that is not a reflection point) is referred to as a blank point.

  The detection image generation unit 30a is configured to periodically generate a detection image for each preset generation cycle SC (see FIG. 12 described later). That is, the detected image generation unit 30a generates a detected image every time when a time set in advance as the generation cycle SC elapses. Since the rain condition changes with the passage of time outdoors, detection images are periodically generated to determine whether it is raining. Note that the timing at which the distance measuring sensor 11 transmits the detection wave is not particularly limited, and the detection wave may always be transmitted, or may be periodically transmitted according to the generation cycle SC. It is good also as a structure which performs ranging. Hereinafter, for the sake of explanation, the time set as the generation cycle SC may be referred to as a frame. For example, the two cycles before the generation cycle SC that generated the detected image may be referred to as two frames before.

  FIG. 3A shows a first detection image GZ1 in which raindrops 52 are detected in a rainy state, and the number of reflection points HK1 is “11”. The reflection points HK1 are discretely located in the first detection image GZ1, and are independent from each other. That is, during the rain, a large number of raindrops 52 exist within the detection range KH, and each of the many raindrops 52 has a small size, and is represented by one pixel.

  FIG. 3B shows the second detection image GZ2 in which the elongated obstacle 51 is detected, and the number of reflection points HK2 is “11”. Since the second detection image GZ2 is not in a rainy state, there are no reflection points due to the raindrops 52. The reflection points HK2 are located side by side in the vertical direction Y, and all the coordinates in the horizontal direction X are the same. In the second detection image GZ2, the rod-shaped obstacle 51 is detected by the distance measuring sensor 11. A block having a large size such as the obstacle 51 is indicated by a plurality of continuous pixels.

  FIG. 3C shows a third detection image GZ3 in which raindrops 52 and obstacles 51 are detected in a rainy state, and the total number of reflection points is “26”. Of the reflection points shown in the third detection image GZ3, the number of reflection points HK3 corresponding to the raindrops 52 is “12”, and the reflection point HK4 corresponding to the obstacle 51 is “14”. The reflection points HK3 corresponding to the raindrops 52 are discretely located in the third detection image GZ3, and the reflection points HK4 corresponding to the obstacle 51 are located in the center of the third detection image GZ3. ing. In FIG. 3C, in order to clarify the difference from the reflection point HK3 indicated by the solid line, the reflection point HK4 is indicated by a thick line, but both are treated as “1” in the image data, and the difference between the two is the difference. There is no.

  The obstacle determination unit 30d determines whether the detected object 50 is an obstacle 51 based on the detected image. Specifically, an obstacle threshold value related to the number of reflection points is set in advance, and the obstacle determination unit 30d determines that the detected object 50 is detected when the number of reflection points in the detection image exceeds the obstacle threshold value. It is determined that the obstacle 51 is present.

  For example, when the reflection point threshold is set to “10”, the first detection image GZ1 to the third detection image GZ3 are all determined that the detected object 50 is the obstacle 51. However, in the first detection image GZ1, only the small raindrops 52 are detected, and it is not preferable to determine that the obstacle 51 is the same as the second detection image GZ2 and the third detection image GZ3. . Therefore, in the obstacle detection device 10, the rain removal processing unit 30c removes the reflection points due to the raindrops 52 from the detected image, thereby avoiding erroneous detection of rain.

  Specifically, the rain removal processing unit 30c performs rain removal processing for removing isolated points from the detected image. In the present embodiment, the rain removal processing uses a median filter (median processing). In general, the median processing is processing for replacing the value of each pixel with the median value of surrounding pixels. Thereby, the isolated point can be reliably removed from the detected image. Hereinafter, the median process will be described with reference to a detection image obtained by performing the median process on the first detection image GZ1 to the third detection image GZ3.

  FIG. 4A is an explanatory diagram illustrating a fourth detection image that has been subjected to rain removal processing by the rain removal processing unit, and FIG. 4B is an explanatory diagram illustrating a fifth detection image that has been subjected to rain removal processing by the rain removal processing unit. .

  FIG. 4A shows a fourth detection image GZ4 obtained by performing median processing on the first detection image GZ1 and the second detection image GZ2, and the number of reflection points is “0”.

  First, in the first detection image GZ1 (see FIG. 3A), the reflection points HK1 are independent of each other and are surrounded by blank points, so that all the reflection points HK1 are replaced with blank points. As a result, the first detection image GZ1 is a fourth detection image GZ4 in which no reflection point exists.

  Next, in the second detection image GZ2 (see FIG. 3B), the reflection points HK2 are continuous, but there are many blank points around the reflection point HK2, and when the median processing is performed, the reflection point HK2 Is replaced with a blank point. As a result, the second detection image GZ2 is a fourth detection image GZ4 that does not have a reflection point, like the first detection image GZ1.

  FIG. 4B shows a fifth detection image GZ5 obtained by performing median processing on the third detection image GZ3, and the number of reflection points HK4 is “14”. In the third detection image GZ3 (see FIG. 3C), there is a reflection point HK3 corresponding to the raindrop 52 and a reflection point HK4 corresponding to the obstacle 51. However, the reflection point HK3 corresponding to the raindrop 52 is the first point. In the same manner as the detected image GZ1, all are replaced with blank points. On the other hand, the reflection point HK4 corresponding to the obstacle 51 is concentrated in position (coordinates), and there are many reflection points HK4 in the surroundings. Therefore, even if median processing is performed, the reflection point HK4 is not replaced with a blank point. It remains as a reflection point HK4. As a result, the third detection image GZ3 is a fifth detection image GZ5 in which only the reflection point HK3 corresponding to the raindrop 52 is removed and only the reflection point HK4 corresponding to the obstacle 51 is present.

  By the way, since the second detection image GZ2 is an image in which the elongated obstacle 51 is detected, it is desirable to determine that the detected object 50 is the obstacle 51. However, when the median process is performed, the second detection image GZ2 is the fourth detection image GZ4. However, when the obstacle determination is performed on the fourth detection image GZ4, there is no reflection point. The object 50 is not determined to be the obstacle 51. Therefore, it is desirable to perform rain removal processing to remove the raindrops 52 during rainfall, but it is desirable not to perform rain removal processing when it is not raining, and to perform rain removal processing depending on the situation. It is desirable to select whether or not. Therefore, in the obstacle detection device 10, the rain determination unit 30b determines whether or not it is a rain state, and instructs whether or not rain removal processing is necessary according to the state.

  Specifically, the rain determination unit 30b determines whether or not the detection range KH is in a rain state based on the presence or absence of discretely located isolated points in the detected object 50 in the detection image. ing. In the present embodiment, the configuration is such that the variance value of the detected object 50 in the detected image is calculated at the time of rainfall determination.

  For example, when performing rainfall determination on the first detection image GZ1, first, median processing is performed on the first detection image GZ1, and the above-described fourth detection image GZ4 (median processing image) is generated.

  Next, a difference image in which a difference between the first detection image GZ1 and the median processed image is calculated is generated. Here, since all the fourth detection images GZ4 are blank points, the difference from the first detection image GZ1 is substantially the same as that of the first detection image GZ1. That is, the difference image in the first detection image GZ1 corresponds to the first detection image GZ1, and the drawing is omitted.

  Then, the variance value in each of the horizontal direction X and the vertical direction Y is calculated for the difference image. In the rain determination, a dispersion threshold regarding the number of dispersion values is set in advance, and the rain determination unit 30b determines that the rain state is present when the calculated dispersion value exceeds the dispersion threshold. In this embodiment, a horizontal dispersion threshold value Thx in the horizontal direction X is set for the horizontal dispersion value Vx in the horizontal direction X, and the vertical direction is set for the vertical dispersion value Vy in the vertical direction Y. A Y vertical dispersion threshold Thy is set. Furthermore, in the rain determination, if “Vx> Thx” and “Vy> Thy”, it is determined that the vehicle is in a rain state. That is, when the variance value in either the horizontal direction X or the vertical direction Y does not exceed the corresponding variance threshold value, it is determined that the rain state is not present (normal state).

  In the case of the first detection image GZ1, since the reflection points HK1 are independent of each other and vary without being collected at one place, the dispersion value becomes a large value exceeding the dispersion threshold. Therefore, it is determined that the first detection image GZ1 is in a rainy state.

  As described above, the rain determination unit 30b performs the rain determination by comparing the detected image with the median processed image from which the isolated point is removed using the median filter. Therefore, even when both the obstacle 51 and the raindrop 52 are detected, the raindrop 52 can be extracted using the median filter, and only the raindrop 52 can be the target of the rain determination.

  By the way, when the rain detection is performed on the second detection image GZ2, the difference image in the second detection image GZ2 is substantially the same as the second detection image GZ2. When the dispersion value in the second detection image GZ2 is calculated, the reflection points HK2 are arranged in the vertical direction Y, and the vertical dispersion value Vy is large because the coordinates in the vertical direction Y are different at each reflection point HK2. Value. On the other hand, in the horizontal direction X, since the coordinates of the reflection point HK2 are the same, the lateral dispersion value Vx is smaller than the lateral dispersion threshold Thx. As a result, it is determined that the second detection image GZ2 is not in a rain state.

  As described above, it is possible to determine whether or not the detected object 50 is an isolated point discretely located by calculating the dispersion value and grasping the variation of the reflection point.

  Next, the processing flow of the obstacle detection method in the obstacle detection apparatus 10 (moving body 1) will be described with reference to the drawings.

  FIG. 5 is a flowchart showing a processing flow of the obstacle detection method of the obstacle detection device according to the first embodiment of the present invention.

  The mobile body 1 is in a state of traveling on the traveling surface 100 at a preset traveling speed at the start of processing.

  In step S101, a detection image is generated by the detection image generation unit 30a. For example, the detected image may be temporarily stored in a storage device or the like, and a plurality of frames of detected images may be accumulated.

  In step S102, a rain determination process (see FIG. 6) described later is performed by the rain determination unit 30b.

  In step S103, it is determined whether or not the rain determination unit 30b determines that the rain state is present. That is, it is determined whether or not the result output from the rain determination unit 30b is in a rain state. As a result, when it is determined that it is raining (step S103: Yes), the process proceeds to step S104. On the other hand, when it is determined that it is not raining (step S103: No), the process proceeds to step S106.

  In step S104, the traveling control unit 30e switches to slow traveling. That is, the traveling control unit 30e instructs the driving unit 20 to reduce the traveling speed. In the travel control unit 30e, a slow speed corresponding to slow travel may be set in advance, or may be a travel speed obtained by subtracting a predetermined value from the normal travel speed. Note that if the vehicle has already been decelerated by slow travel, there is no need to change the setting.

  In step S105, a rain removal process (see FIG. 7) described later is performed by the rain removal processing unit 30c, and the process proceeds to step S108.

  In step S106, the travel control unit 30e switches to normal travel. That is, when the traveling speed of the moving body 1 is decelerated like the slow speed, it returns to the traveling speed at the start of the process. Moreover, when it is set as normal driving | running and it is set as the initial driving speed, it is not necessary to change a setting.

  In step S107, the rain removal processing unit 30c stops the rain removal processing. That is, if rain removal processing has been performed in the previous frame, the rain removal processing is stopped, and the detection image generated in step S101 is output without performing rain removal processing.

  In step S108, the obstacle determination unit 30d performs an obstacle determination process. In the obstacle determination process, the obstacle determination is performed on the detected image. However, when the rain removal process is performed in step S105, the obstacle determination is performed on the processed image subjected to the rain removal process.

  In step S109, the obstacle determination unit 30d determines whether it is determined that there is the obstacle 51. That is, the determination is made based on whether or not the detected object 50 in the detected image is the obstacle 51. As a result, when it is determined that there is no obstacle 51 (step S109: No), the process proceeds to step S110. On the other hand, if it is determined that there is an obstacle 51 (step S109: Yes), the process proceeds to step S111.

  In step S110, the detected image generation unit 30a waits for the next generation cycle, and then returns to step S101. That is, the obstacle detection device 10 waits for one frame to elapse and repeats the processes shown in steps S101 to S109. Note that the moving body 1 continues to travel without waiting while one frame elapses. Therefore, the obstacle detection device 10 periodically detects an obstacle while the moving body 1 is traveling. In addition, when the traveling speed is changed during the process, the changed speed is maintained until it is changed again.

  In step S111, the traveling control unit 30e stops traveling and ends the process. That is, when it is determined that the obstacle 51 is within the detection range KH, the moving body 1 stops and waits for an instruction.

  Next, the rain determination process in step S102 will be described in detail. In the present invention, a plurality of methods can be used as rain determination and rain removal processing. Therefore, in order to distinguish between a plurality of rain determination processes and rain removal processes, the rain determination process and the rain removal process in the first embodiment of the present invention are referred to as a rain determination process 1 and a rain removal process 1.

  FIG. 6 is a flowchart showing a process flow of the rain determination process 1 in the first embodiment of the present invention.

  In step S201, the median processing image is generated by the rain determination unit 30b. That is, the median process is performed on the detected image generated in step S101 to obtain a median processed image.

  In step S202, the rain determination unit 30b generates a difference image that compares the detected image with the median processed image.

  In step S203, the rain determining unit 30b determines whether or not the variance value of the difference image exceeds the variance threshold. That is, the lateral dispersion value Vx and the longitudinal dispersion value Vy in the difference image are calculated and compared with the lateral dispersion threshold Thx and the longitudinal dispersion threshold Thy. As a result, when the variance value exceeds the variance threshold (step S203: Yes), the process proceeds to step S204. On the other hand, if the variance value does not exceed the variance threshold (step S203: No), the process proceeds to step S205.

  When the number of reflection points does not exceed the reflection point threshold, compared to the reflection point threshold set in advance for the number of reflection points in the difference image before comparing the dispersion value and the dispersion threshold. However, the process may proceed to step S205 regardless of the variance value. That is, if the number of reflection points is small, the accuracy of the dispersion value is low, and it is difficult to determine whether the reflection points actually vary. Moreover, if there are few reflection points, even if it is raining, it can be estimated that it is not heavy rain that requires rain removal processing.

  In step S204, the rain determination unit 30b determines that it is in a rain state, and the rain determination process 1 ends.

  In step S205, the rain determination unit 30b determines that the rain state is not present, and the rain determination process 1 is terminated.

  FIG. 7 is a flowchart showing the process flow of the rain removal process 1 in the first embodiment of the present invention.

  In step S301, the rain removal processing unit 30c performs median processing on the detected image. Note that if a median-processed image is generated in the above-described rainfall determination process (for example, step S201), it can be used, and it is not necessary to generate a new image, so that the processing time can be shortened. it can.

  In step S302, the rain removal processing unit 30c designates the detected image that has undergone the median process as a processed image, and the process ends. That is, the above-described obstacle determination is performed on the processed image designated here.

  In the present embodiment, the above-described processing is performed, and hereinafter, as specific examples, cases where the first detection image GZ1 to the third detection image GZ3 are generated in step S101 will be described.

  When the first detection image GZ1 is generated, it is determined in the rain determination process in step S102 that it is in a rain state (step 204 in FIG. 6). As a result, the moving body 1 is set to travel slowly, and rain removal processing (step 105 in FIG. 5) is performed on the first detection image GZ1. As a result of the rain removal process, the first detection image GZ1 is the fourth detection image GZ4 from which the reflection point HK1 has been removed, and obstacle determination is performed on the fourth detection image GZ4. Since it is determined that the fourth detected image GZ4 has no obstacle 51, the process proceeds to step S110. Thereafter, the moving body 1 travels slowly, and a detection image is generated again, and the obstacle 51 is detected.

  When the second detection image GZ2 is generated, it is determined in the rain determination process that the rain state is not present (step 205 in FIG. 6). As a result, the moving body 1 is in a normal travel, and no rain removal process is performed. In the obstacle determination, it is determined that there is an obstacle 51 for the second detection image GZ2 (step S111 in FIG. 5). Then, the moving body 1 stops traveling.

  When the third detection image GZ3 is generated, it is determined that it is raining in the rain determination process. The third detection image GZ3 is subjected to rain removal processing, and the obstacle determination target is the fifth detection image GZ5. In the obstacle determination, it is determined that there is an obstacle 51 and the moving body 1 is stopped.

  As described above, in the distance measurement result of the distance measurement sensor 11, both the raindrop 52 and the obstacle 51 are regarded as the detected object 50 and cannot be distinguished from each other. Therefore, by performing rain removal processing on the detection image during the rain, a detection image showing only the obstacle 51 can be obtained, and erroneous detection of the raindrop 52 as the obstacle 51 can be avoided.

  In addition, the traveling control unit 30e is configured to control to reduce the traveling speed when the rain determining unit 30b determines that it is in a raining state. Accordingly, in consideration of the fact that the time required for the processing becomes longer due to the rain removal processing or the like during the rain, the moving body 1 is decelerated to increase the time until it approaches the obstacle 51 and the like. Collision with 51 etc. can be avoided. For example, when a slender obstacle 51 as shown in FIG. 3B exists in a rainy state, the obstacle 51 may be removed by the rain removal process. Obstacle determination is performed again before approaching 51. As a result, by detecting after approaching, the area occupied by the obstacle 51 in the detection image becomes large, and it can be determined that the obstacle 51 is not removed by the rain removal process and the obstacle 51 is present.

  The obstacle detection method according to the present invention transmits a detection wave to the detected object 50, receives a reflected wave from the detected object 50, and measures a detection distance KL to the detected object 50. An obstacle detection method in the obstacle detection apparatus 10 provided with the sensor 11, wherein the detected image is generated in the detection range KH where the detection wave is transmitted to the detection image generation unit 30 a based on the distance measurement result of the distance measurement sensor 11. Based on the detection image generation step for generating a detection image indicating the presence or absence of the detection object 50 and the presence / absence of isolated points discretely located among the detection objects 50 in the detection image in the rain determination unit 30b. A rain determination step for determining whether or not KH is in a rain state, a rain removal processing step for causing the rain removal processing unit 30c to perform a rain removal process for removing isolated points from the detected image, and an obstacle determination unit 30d, detected image Based on, and a obstacle determining step of the obstacle determines whether the detection object 50 is an obstacle 51. In the obstacle determination step, when it is determined in the rain determination step that the vehicle is in the rain state, the obstacle determination is performed based on the detected image subjected to the rain removal process in the rain removal processing step.

(Second Embodiment)
Next, an obstacle detection device and a moving body according to a second embodiment of the present invention will be described with reference to the drawings. Since the second embodiment has substantially the same shape as the first embodiment, the external view and the configuration diagram are omitted, and the components having substantially the same functions as those of the first embodiment are omitted. The same reference numerals are given and the description is omitted.

  FIG. 8 is an explanatory diagram illustrating a sixth detection image generated by the detection image generation unit according to the second embodiment of the present invention.

  The second embodiment differs from the first embodiment in the format of the detection image generated by the detection image generation unit 30a. Specifically, the detected image is a two-dimensional planar image in the first embodiment, whereas in the second embodiment, the detected image is a three-dimensional stereoscopic image. Yes. That is, the detection image generation unit 30a generates a three-dimensional image (for example, the sixth detection image GZ6) corresponding to the detection range that is a three-dimensional space as the detection image.

  FIG. 8 is the sixth detection image GZ6 generated in the second embodiment, and is an image in which the reflection point HK5 is arranged in a three-dimensional space. In the sixth detection image GZ6, in addition to the horizontal direction X and the vertical direction Y, a depth direction Z orthogonal to both is shown. When the reflection point HK5 is arranged, the position in the depth direction Z may be determined based on the detection distance KL of the detection object 50. The sixth detection image GZ6 shows the raindrop 52 detected during the rain as the reflection point HK5, and the reflection point HK5 is separated in the horizontal direction X and the vertical direction Y, respectively, in addition to the depth direction Z. But it is far away.

  In the second embodiment, the same obstacle detection method as that in the first embodiment is implemented. In the rain determination, the variance value is calculated in the depth direction Z along with the horizontal direction X and the vertical direction Y with respect to the sixth detection image GZ6. If the calculated dispersion value does not exceed the corresponding dispersion threshold value in any one of the horizontal direction X, the vertical direction Y, and the depth direction Z, it is determined that the rain state is not present (normal state). Is done. Therefore, it is possible to determine whether or not it is raining more reliably by performing the process of detecting rain on the three-dimensional space. That is, even when a plurality of reflection points HK5 are adjacent to each other and have a shape like a large obstacle 51, it is grasped that they are separated from each other by comparing the distances to the respective reflection points HK5. it can.

  When generating the sixth detection image GZ6, the detection image generation unit 30a may generate not only a three-dimensional image but also an image (for example, the first detection image GZ1) showing the detection result on a plane. The rain removal process and the obstacle determination are performed on a planar detection image, thereby simplifying the image process and shortening the processing time. In addition, when an obstacle is determined for a three-dimensional image, the shape of the detected object 50 can be grasped more accurately, so that the determination accuracy can be improved.

(Third embodiment)
Next, an obstacle detection device and a moving body according to a third embodiment of the present invention will be described with reference to the drawings. Since the third embodiment is substantially the same as the first embodiment, the external view and the configuration diagram are omitted, and the same components are substantially the same in function as the first embodiment and the second embodiment. The description will be omitted.

  FIG. 9 is an explanatory diagram illustrating a seventh detection image generated by the detection image generation unit according to the third embodiment of the present invention, and FIG. 10 illustrates a difference between the first detection image and the seventh detection image. It is explanatory drawing which shows an image.

  The third embodiment differs from the first embodiment in the method used for rainfall determination. In the first embodiment, the difference between the detected image and the median-processed image obtained by performing the median process is calculated. However, in the third embodiment, the rainfall determination unit 30b is generated at different generation cycles SC. It is set as the structure which compares two detection images and performs rain judgment.

  FIG. 9 shows a seventh detection image GZ7 in which raindrops 52 are detected in a rainy state, and the number of reflection points HK6 is “10”. The reflection points HK6 are discretely located in the seventh detection image GZ7 and are independent from each other. The seventh detection image GZ7 is a detection image generated one frame before the first detection image GZ1. That is, since the installed obstacle 51 does not move, the position does not change even if time elapses. However, since the raindrop 52 moves as time elapses, the position in the detection range KH changes. Therefore, the seventh detection image GZ7 is different from the first detection image GZ1 in the arrangement of the reflection points HK6 (raindrops 52). In FIGS. 9 and 10, in order to clarify the difference from the reflection point HK1 indicated by the solid line, the reflection point HK6 is indicated by a broken line, but both are treated as “1” in the image data. There is no difference between the two.

  FIG. 10 shows an eighth detection image GZ8 (difference image) obtained by calculating the difference between the first detection image GZ1 and the seventh detection image GZ7. The reflection point HK1 of the first detection image GZ1 and the seventh detection image GZ7. The reflection point HK6 is shown. In the difference image, the overlapping reflection points are removed. In the third embodiment, the rain determination is performed on the eighth detection image GZ8, and it is determined whether it is raining based on the variance value in the eighth detection image GZ8. Thus, by comparing the detection images having different generation cycles SC, it is possible to grasp the raindrops that have moved over time. In addition, since image processing or the like is not performed on the detected image, it is possible to perform rainfall determination without removing a small obstacle or the like.

  Next, a processing flow of the obstacle detection method in the third embodiment will be described with reference to the drawings. In the third embodiment, the rain determination process is different from the first embodiment, but the other processes are the same. That is, in the third embodiment, the above-described process flow shown in FIG. 5 is performed, and the rain determination process 2 shown below is performed in step S102.

  FIG. 11 is a flowchart showing the process flow of the rain determination process 2 in the third embodiment of the present invention.

  In step S401, the rain detection unit 30b designates the detection image generated in the previous generation cycle as a comparative detection image. That is, the detection image generated in step S101 one frame before is set as a comparative detection image.

  In step S402, the rain determination unit 30b generates a difference image that compares the detected image with the comparatively detected image. That is, when the first detection image GZ1 is a detection image and the seventh detection image GZ7 is a comparison detection image, an eighth detection image GZ8 is generated as a difference image.

  Since the processing from step S403 to step S405 is the same as that from step S203 to step S205 described above, description thereof will be omitted.

(Fourth embodiment)
Next, an obstacle detection device and a moving body according to a fourth embodiment of the present invention will be described with reference to the drawings. Since the fourth embodiment is substantially the same as the first embodiment, the external view and the configuration diagram are omitted, and the same components are substantially the same in function as the first to third embodiments. The description will be omitted.

  FIG. 12 is an explanatory diagram showing the relationship between the generation period and the rain determination period in the fourth embodiment of the present invention.

  The fourth embodiment is different from the first embodiment in the period for performing rainfall determination. In the first embodiment, the rain determination is performed for each generation cycle SC in which the detection image is generated. In the fourth embodiment, the rain determination is performed for each of the plurality of generation cycles SC.

  FIG. 12 shows the timing of performing rainfall determination as time elapses. In FIG. 12, time T1 to time T11 indicate the time when the detected image is generated. For example, the interval between time T1 and time T2 corresponds to one generation cycle SC (one frame).

  Explaining along the passage of time, first, at time T1, a detection image is generated, and rainfall determination is performed. Next, at time T2 and time T3, a detection image is generated, but rain determination is not performed. Then, at time T4, the rain detection is made while the detection image is generated. Thereafter, the rain determination is performed only at time T7 and time T10, and is not performed in other frames. Thus, while the detection image is generated every frame, the rain determination is performed every three frames, and the same processing is repeated after time T10.

  As described above, the rain determination unit 30b performs the rain determination for each of the plurality of generation cycles SC, and the interval for performing the rain determination is set in advance as the rain determination cycle KC. In the present embodiment, the rain determination period KC is set to 3 frames. However, the present invention is not limited to this, and the rain determination interval may be changed as appropriate. As described above, by performing the rain determination after a certain period of time, an increase in processing time due to the rain determination can be minimized. That is, in the obstacle detection device 10, the generation cycle SC is set to be short so that the obstacle 51 can be quickly dealt with, but the change in weather is gentle compared to the obstacle 51 and the like. For this reason, the frequency of rain determination may be lowered to suppress excessive processing.

  Next, a processing flow of the obstacle detection method in the fourth embodiment will be described with reference to the drawings. In the fourth embodiment, substantially the same processing as in the first embodiment is performed, so that different points will be described in detail, and descriptions of other points will be omitted.

  FIG. 13: is a flowchart which shows the processing flow of the obstruction detection method of the obstruction detection apparatus concerning 4th Embodiment of this invention.

  Similar to the first embodiment, the moving body 1 is in a state of traveling on the traveling surface 100 at the traveling speed at the start of processing.

  In step S501, similarly to the first embodiment, a detection image is generated by the detection image generation unit 30a.

  In step S502, the rain determination unit 30b determines whether or not the rain determination period KC has been reached. Here, it is determined whether or not the time set as the rainfall determination period KC has elapsed since the previous rain determination was performed. As a result, when the rainfall determination period KC is reached (step S502: Yes), the process proceeds to step S503. On the other hand, if the rainfall determination period KC has not been reached (step S502: No), the process proceeds to step S509.

  The processing from step S503 to step S512 is the same as that from step S102 to step S111 described above, and a specific description thereof will be omitted. That is, when the rainfall determination period KC is reached, the rain determination process is performed in step S503 corresponding to step S102. On the other hand, if the rainfall determination period KC has not been reached, an obstacle determination process is performed in step S509 corresponding to step S108. Therefore, whether or not the rain determination is necessary is determined depending on whether or not the rain determination cycle KC has been reached. In the rain determination process in step S503, the rain determination process 1 shown in FIG. 6 may be applied, or the rain determination process 2 shown in FIG. 11 may be applied.

(Fifth embodiment)
Next, an obstacle detection device and a moving body according to a fifth embodiment of the present invention will be described with reference to the drawings. Since the fifth embodiment is substantially the same as the first embodiment, the external view and the configuration diagram are omitted, and the components that have substantially the same functions as those of the first to fourth embodiments are the same. The description will be omitted.

  FIG. 14 is an explanatory diagram showing a relationship between repetition of rainfall determination and output of determination results in the fifth embodiment of the present invention.

  The fifth embodiment is different from the fourth embodiment in that the determination result is output after the rain determination is repeated. In the fourth embodiment, the determination result of the rain determination is immediately output. In the fifth embodiment, the determination result is output when the same determination result is continued.

  FIG. 14 shows the timing for performing the rain determination as time elapses, as in FIG. In FIG. 14, as in FIG. 12, the rain determination is performed every three frames, and the rain determination is performed at time T <b> 1, time T <b> 4, time T <b> 7, and time T <b> 10. As a result of the rain determination, it is determined that it is not raining at time T1, and it is determined that it is raining at time T4, time T7, and time T10. In the present embodiment, the determination result is output when the same determination is repeated three times. At time T10, the determination result indicating that it is in a rainy state is output. Specific processing will be described in conjunction with the processing flows shown in FIGS. 15 and 16.

  Next, a processing flow of the obstacle detection method in the fifth embodiment will be described with reference to the drawings. In 5th Embodiment, since the process substantially the same as 4th Embodiment is performed, a different point is demonstrated in detail and description is abbreviate | omitted about another point.

  FIG. 15 is a flowchart showing a process flow of the obstacle detection method of the obstacle detection apparatus according to the fifth embodiment of the present invention, and FIG. 16 is a process of the continuous determination process in the fifth embodiment of the present invention. It is a flowchart which shows a flow.

  Steps S601 to S603 are the same as steps S501 to S503 described above, and steps S605 to S613 are similar to steps S504 to S512 described above. Is omitted. That is, the fifth embodiment is different from the fourth embodiment in that the continuous determination process shown in FIG. 16 is performed as step S604 after step S603. Therefore, the continuous determination process will be described with reference to FIG.

  In step S701, the rain determination unit 30b determines whether the determination result in the rain determination process (step S603) is the same as the determination result in the previous rain determination process. For example, if it is rain determination at time T4 in FIG. 14, the determination result is compared with the rain determination at time T1. As a result, when the determination result is the same as the previous time (step S701: Yes), the process proceeds to step S702. On the other hand, if the determination result is different from the previous time (step S701: No), the process proceeds to step S703.

  In step S702, the number of determinations is increased by one by the rain determination unit 30b. In the present embodiment, the result of rain determination is counted and stored as the number of determinations. For example, when the number of determinations is stored as one, it is stored as two through step S702.

  In step S703, the number of determinations is reset by the rain determination unit 30b. Here, regardless of the stored number of determinations, the number of determinations is stored as one.

  In step S704, the rain determination unit 30b determines whether the number of determinations exceeds a specified value. The specified value is set to 3 times in advance, and when the number of determinations is 3, it is determined that the specified value has been exceeded. As a result, when the number of determinations exceeds the specified value (step S704: Yes), the process proceeds to step S705. On the other hand, when the number of determinations does not exceed the specified value (step S704: No), the process proceeds to step S706. In the present embodiment, the specified value is set to 3 times, but the specified value may be set as appropriate.

  In step S705, the rain determination unit 30b outputs the determination result in the current generation cycle SC, and the process ends.

  In step S706, the determination result output last time is output by the rain determination unit 30b, and the process ends.

  After the continuous determination process, in step S605, the rain determination unit 30b determines whether the output determination result is in a rain state. In other words, in step S605, the same processing as in step S504 is performed. If it is raining, the process proceeds to step S606, and measures such as slow running of the moving body 1 are taken. On the other hand, if it is not raining, the process proceeds to step S608, and the moving body 1 is caused to travel normally.

  In the present embodiment, the above-described processing is performed, and in the following, the continuous determination processing will be described by taking a case illustrated in FIG. 14 as a specific example. In addition, before time T1, it is assumed that the determination result that it is not raining is continuously output.

  At time T1, it is determined that it is not raining, and the determination result is the same as the previous time, so the determination count is increased by one (step S702). And the determination result that the frequency | count of determination exceeds the regulation value and it is not a rainy state is output (step S705).

  At time T4, it is determined that it is raining and is different from the previous determination result, so the determination count is reset (step S703). Then, since the number of times of determination is one and does not exceed the specified value, the same determination result (not in the rain state) as the time T1 is output (step S706).

  At time T7, it is determined that it is raining, and the determination result is the same as the previous time, so the number of determinations is increased by 1 to 2 times. Since the number of determinations does not exceed the specified value, the same determination result as that at time T4 is output.

  At time T10, it is determined that it is raining, and since the determination result is the same as the previous time, the number of determinations is increased by 1 to 3 times. As a result, the number of determinations exceeds the specified value, and the determination result at time T10 (in a rainy state) is output (step S705).

  As described above, it is determined that it is raining at time T4. However, since it is determined that it is not raining at the previous time T1, the same determination is not continuous and it is raining. Is not output. Then, although it is determined that it is raining at time T7, the determination result is not output because the number of continuous determinations is two. Furthermore, when it is determined that it is raining at time T10, the number of consecutive determinations is 3, and a determination result indicating that it is raining is output. When the rain judgment is repeated, rain or the like continues to some extent, so the same result continues, but if it is temporary noise, the result will be different. When the same result continues, it is possible to prevent erroneous detection of noise or the like in the detected image as raindrops by determining whether or not it is raining.

  In the fifth embodiment, the rain determination is performed for each of a plurality of frames. However, the present invention is not limited to this, and the rain determination may be performed for each frame as in the first embodiment.

(Sixth embodiment)
Next, an obstacle detection device and a moving body according to a sixth embodiment of the present invention will be described with reference to the drawings. Since the sixth embodiment is substantially the same as the first embodiment, the external view and the configuration diagram are omitted, and the same components are substantially the same in function as the first to fifth embodiments. The description will be omitted.

  FIG. 17A is an explanatory diagram of first distance data indicating a detection distance, and FIG. 17B is an explanatory diagram of second distance data indicating a detection distance. In FIG. 17A and FIG. 17B, hatching is performed to emphasize a part.

  In the sixth embodiment, the rain determination unit 30b is configured to calculate a difference in the detection distance KL from the surroundings of the detected object 50 in the detection image at the time of the rain determination. In the rain determination, a measurement point (pixel) having a large difference in the detection distance KL from the surroundings is set as an isolated point, and it is determined whether it is raining based on the number of isolated points.

  In the present embodiment, when generating a detection image, distance data indicating a detection distance KL for each coordinate is generated. The first distance data DT1 and the second distance data DT2 shown in FIGS. 17A and 17B are specific examples in which a part of the generated distance data is enlarged. In the distance data, coordinates are indicated by a matrix, and the coordinates and the detection distance KL are associated with each other. The first distance data DT1 and the second distance data DT2 are “3 × 3” matrices. In the horizontal direction X, “1 column”, “2 columns”, and “3 columns” are arranged, and in the vertical direction Y, “A row”, “B row”, and “C row” are arranged. In the following, for simplification of description, coordinates are shown together with rows and columns. For example, if the coordinates are in the upper left (first row, first column), they are called “A1”, and if the coordinates are in the middle right (second row, third column), they are called “B3”. The first distance data DT1 and the second distance data DT2 are enlarged views of a part of the distance data, and the distance data is composed of more rows and columns than “3 × 3”. Also good.

  As for the detection distance KL in the first distance data DT1 shown in FIG. 17A, “B2” is 1 m (meters), and other coordinates are 10 to 13 m. Specifically, a distance threshold relating to the difference in the detection distance KL is set in advance, and in this embodiment, the distance threshold is 5 m. Then, when the detection distance KL between “B2” and the surrounding coordinates is compared, the difference between the detection distances KL exceeds the distance threshold value, so that “B2” is determined to be an isolated point.

  The detection distance KL in the second distance data DT2 shown in FIG. 17B is “B2” is 1 m, the detection distance KL of “B1” adjacent to “B2” is 4 m, and other coordinates are 10 to 13 m. It is. In the second distance data DT2, since the difference in the detection distance KL between “B2” and “B1” does not exceed the distance threshold, “B2” and “B1” are determined not to be isolated points.

  FIG. 18A is an explanatory diagram illustrating a ninth detection image generated by the detection image generation unit, and FIG. 18B is an explanatory diagram illustrating a tenth detection image generated by the detection image generation unit.

  The detection image generation unit 30a generates a detection image that shows the isolated points described above as reflection points. That is, in the image data, the pixel corresponding to the isolated point is indicated as “1”, and the other pixels are indicated as “0”. Specifically, in the ninth detection image GZ9 shown in FIG. 18A, the reflection points HK7 are discretely located in the ninth detection image GZ9, and the number is “13”. In the tenth detection image GZ10 shown in FIG. 18B, the reflection points HK8 are discretely located in the tenth detection image GZ10, and the number is “4”. In the rain determination, an isolated point threshold relating to the number of isolated points is set in advance, and in this embodiment, the isolated point threshold is set to “10”. That is, since the number of reflection points HK7 (isolated points) exceeds the isolated point threshold, the ninth detection image GZ9 is determined to be in a rain state, and the tenth detection image GZ10 has the number of reflection points HK8 isolated. Since the point threshold value is not exceeded, it is determined that it is not raining. Note that the distance threshold and the isolated point threshold may be set as appropriate.

  Next, a processing flow of the obstacle detection method in the sixth embodiment will be described with reference to the drawings. In the sixth embodiment, the rain determination process is different from the first embodiment, but the other processes are the same. That is, in the sixth embodiment, the above-described process flow shown in FIG. 5 is performed, and the rain determination process 3 shown below is performed in step S102.

  FIG. 19 is a flowchart showing a process flow of the rain determination process 3 in the third embodiment of the present invention.

  In step S801, distance data (for example, first distance data DT1 and second distance data DT2) indicating the detection distance KL for each detection object 50 is generated by the detection image generation unit 30a.

  In step S802, the rain determination unit 30b counts the number of isolated points in the distance data. An isolated point is determined based on a distance threshold. Here, the isolated points in the distance data may be counted, or the reflection points of the detected image corresponding to the distance data may be counted.

  In step S803, the rain determination unit 30b determines whether the number of isolated points exceeds the isolated point threshold. As a result, when the number of isolated points exceeds the isolated point threshold (step S803: Yes), the process proceeds to step S804. On the other hand, if the number of isolated points does not exceed the isolated point threshold (step S803: No), the process proceeds to step S805.

  In step S804, the rain determination unit 30b determines that it is in a rain state, and the rain determination process 3 ends.

  In step S805, the rain determination unit 30b determines that it is not in a rain state, and the rain determination process 3 ends.

  As described above, it is possible to determine whether or not the raindrops 52 are discretely located by making a comparison based on the detection distance KL. Further, even when the detected object 50 is locally concentrated due to the presence of the obstacle 51, the difference in the detection distance KL can be grasped without being affected by the statistical bias.

(Seventh embodiment)
Next, an obstacle detection device and a moving body according to a seventh embodiment of the present invention will be described with reference to the drawings. Since the seventh embodiment is substantially the same as the first embodiment, the external view and the configuration diagram are omitted, and the same components are substantially the same in function as the first to sixth embodiments. The description will be omitted.

  FIG. 20A is an explanatory diagram illustrating third distance data in the first frame, FIG. 20B is an explanatory diagram illustrating fourth distance data in the second frame, and FIG. 20C is a diagram illustrating the third frame. FIG. 20D is an explanatory diagram illustrating the sixth distance data obtained from the maximum detected distance. In FIGS. 20A to 20D, hatching is performed to emphasize a part.

  The seventh embodiment is different from the first embodiment in the method used for the rain removal process. In 7th Embodiment, the rain removal process part 30c is set as the structure which compares the detection distance KL for every some production | generation periods SC with respect to a detection image, and removes an isolated point. In the present embodiment, the detection distance KL is compared with three frames including the previous frame and the previous two frames with respect to the frame in which the detection image is generated. In the following, for the sake of explanation, three consecutive frames are referred to as a first frame, a second frame, and a third frame over time.

  The third distance data DT3, the fourth distance data DT4, and the fifth distance data DT5 are part of the distance data corresponding to the detected images in the first frame, the second frame, and the third frame, respectively. Similar to the first distance data DT1, it is a “3 × 3” matrix. The third distance data DT3, the fourth distance data DT4, and the fifth distance data DT5 indicate the same coordinates in the detection range KH.

  The detection distance KL of the third distance data DT3 is “A1” is 2 m, “A2” is 10 m, “A3” is 9 m, “B1” is 13 m, “B2” is 10 m, “B3” is 2 m, “C1”. Is 12 m, “C2” is 4 m, and “C3” is 9 m.

  The detection distance KL of the fourth distance data DT4 is 8m for "A1", 1m for "A2", 6m for "A3", 4m for "B1", 2m for "B2", 2m for "B3", "C1" Is 13 m, “C2” is 6 m, and “C3” is 3 m.

  The detection distance KL of the fifth distance data DT5 is “A1” 7 m, “A2” 15 m, “A3” 3 m, “B1” 4 m, “B2” 10 m, “B3” 2 m, “C1”. Is 8 m, “C2” is 11 m, and “C3” is 5 m.

  In the rain removal process, the maximum detection distance KL at each measurement point is acquired, and maximum value data indicating the maximum value for each measurement point is generated. For example, in the case of “A1”, 8 m of the fourth distance data DT4 is acquired as the maximum value. Similarly, the maximum value is acquired for the other coordinates, and the sixth distance data DT6 shown in FIG. 20D is generated as the maximum value data of the third distance data DT3, the fourth distance data DT4, and the fifth distance data DT5.

  The detection distance KL of the sixth distance data DT6 is 8m for "A1", 15m for "A2", 9m for "A3", 13m for "B1", 10m for "B2", 2m for "B3", "C1" Is 13 m, “C2” is 11 m, and “C3” is 9 m.

  When obstacle determination is performed on the maximum value data generated in this way, it is only necessary to determine whether or not it is a reflection point with reference to a threshold set in advance with respect to the detection distance KL. For example, when the threshold is 5 m, it is determined that “B3” is a reflection point corresponding to the detected object 50 in the sixth distance data DT6. Then, the number of reflection points in the distance data is compared with the obstacle threshold value, and it is determined that the detected object 50 is the obstacle 51.

  Regarding the detection distance of the third distance data DT3 to the fifth distance data DT5, coordinates other than “B3” are also present, and coordinates smaller than the threshold exist, and it is not possible to determine either the obstacle 51 or the raindrop 52. However, the detection distance KL varies compared to other frames. On the other hand, in “B3”, it is 2 m in any frame, and the detection distance KL does not change. In this way, by comparing the detection distance KL for each generation cycle SC, it is possible to distinguish between those that are located at the same place and those that move at high speed like the raindrops 52. For the coordinates where raindrops 52 and the like exist, the noise that hinders obstacle detection can be removed by obtaining the maximum value of the detection distance KL. In addition, since the processing is performed using a simple calculation such as comparing the detection distance KL, the processing speed can be increased.

  Next, a processing flow of the obstacle detection method in the seventh embodiment will be described with reference to the drawings. In the seventh embodiment, the rain removal determination process is different from the first embodiment, but the other processes are the same. That is, in the seventh embodiment, the above-described processing flow shown in FIG. 5 is performed, and in step S105, the rain removal processing 2 shown below is performed.

  FIG. 21 is a flowchart showing the process flow of the rain removal process 2 in the seventh embodiment of the present invention.

  In step S901, the detected image generation unit 30a generates distance data (for example, third distance data DT3 to fifth distance data DT5) indicating the detection distance KL for each detected object 50. In addition, it is good also as a structure which produces | generates only the distance data in the flame | frame which implements rain removal processing, and memorize | stores the produced | generated distance data. That is, the distance data in the previous frame may be referred to the stored distance data.

  In step S902, the rain removal processing unit 30c acquires the maximum value for each measurement point among the distance data in the plurality of generation cycles SC. In the present embodiment, the maximum value in three frames is obtained. However, the present invention is not limited to this, and the number of frames to be compared may be set as appropriate.

  In step S903, the rain removal processing unit 30c generates maximum value data (for example, sixth distance data DT6) indicating the maximum value for each measurement point, and the process ends. Subsequent obstacle determination is performed on the maximum value data.

  In the sixth embodiment and the seventh embodiment, the processing flow shown in FIG. 5 is implemented. However, the present invention is not limited to this, and the processing flow shown in FIG. It is good also as a structure which performs determination.

  It should be noted that the embodiment disclosed herein is illustrative in all respects and does not serve as a basis for limited interpretation. Therefore, the technical scope of the present invention is not interpreted only by the above-described embodiment, but is defined based on the description of the scope of claims. Moreover, all the changes within the meaning and range equivalent to a claim are included.

DESCRIPTION OF SYMBOLS 1 Moving body 10 Obstacle detection apparatus 11 Distance sensor 20 Drive part 30 CPU
30a Detected image generation unit 30b Rain determination unit 30c Rain removal processing unit 30d Obstacle determination unit 30e Travel control unit 50 Detected object 51 Obstacle 52 Raindrop 100 Traveling surface KL Detection distance KH Detection range

Claims (15)

A ranging sensor that transmits a detection wave to the object to be detected, receives a reflected wave from the object to be detected, and measures a detection distance to the object to be detected;
A detection image generation unit that generates a detection image indicating the presence or absence of the detection object in a detection range in which the detection wave is transmitted based on a distance measurement result of the distance sensor;
A rain determining unit that determines whether or not the detection range is in a rain state based on the presence or absence of discrete points located discretely among the detected objects in the detection image;
A rain removal processing unit for performing rain removal processing to remove the isolated points from the detected image;
An obstacle determination unit that determines whether or not the detected object is an obstacle based on the detected image;
The obstacle determination unit performs the obstacle determination based on the detection image subjected to the rain removal processing by the rain removal processing unit when the rain determination unit determines that the rain state is present. Obstacle detection device characterized. The obstacle detection device according to claim 1,
The detection range is a three-dimensional space,
The detection image generation unit generates a three-dimensional image corresponding to a detection range as a three-dimensional space as the detection image,
The obstacle detection device, wherein the rain determination unit performs the rain determination on the three-dimensional image. The obstacle detection device according to claim 1 or 2,
The obstacle detection device, wherein the rain determination unit calculates a variance value of the detected object in the detection image at the time of the rain determination. The obstacle detection device according to claim 1 or 2,
The obstacle detection device, wherein the rain determination unit calculates a difference in the detection distance from the periphery of the detected object in the detection image at the time of the rain determination. The obstacle detection device according to any one of claims 1 to 4, wherein
The obstacle detection device, wherein the detection image generation unit periodically generates the detection image every preset generation cycle. The obstacle detection device according to any one of claims 1 to 5, wherein
The said rain removal process part performs the said rain removal process using a median filter. The obstacle detection apparatus characterized by the above-mentioned. The obstacle detection device according to any one of claims 1 to 6,
The obstacle determination unit compares the detected image with a median-processed image from which the isolated point is removed using a median filter with respect to the detected image, and performs the rainfall determination. . The obstacle detection device according to claim 5,
The obstacle detection device, wherein the rain determination unit performs the rain determination by comparing two detection images generated in different generation cycles. The obstacle detection device according to claim 5 or claim 8,
The obstacle detection device, wherein the rain determination unit performs the rain determination for each of the plurality of generation periods. The obstacle detection device according to claim 5, claim 8, or claim 9,
The obstacle detection device, wherein the rain determination unit performs the rain determination in a plurality of the generation cycles, and outputs the result of the rain determination when the same determination is made a plurality of times in succession. The obstacle detection device according to claim 5, claim 8, claim 9, or claim 10,
The said rain removal process part removes the said isolated point by comparing the said detection distance for every said several production | generation period with respect to the said detection image.   A moving body comprising the obstacle detection device according to any one of claims 1 to 11 and traveling on a traveling surface. The moving body according to claim 12,
A travel control unit for controlling a travel speed of traveling on the travel surface;
The travel control unit performs control to decelerate the travel speed when the rain determination unit determines that the rain state is present. Obstacle detection in an obstacle detection device having a distance measuring sensor that transmits a detection wave to an object to be detected, receives a reflected wave from the object to be detected, and measures a detection distance to the object to be detected A method,
A detection image generation step of causing the detection image generation unit to generate a detection image indicating the presence or absence of the detection object in the detection range in which the detection wave is transmitted, based on a distance measurement result of the distance sensor;
A rainfall determination step for causing a rainfall determination unit to determine whether or not the detection range is in a rainfall state based on the presence or absence of discrete points located discretely among the detected objects in the detection image; ,
A rain removal processing step for causing a rain removal processing unit to perform rain removal processing for removing the isolated points from the detected image;
An obstacle determination step for causing the obstacle determination unit to determine whether the detected object is an obstacle based on the detected image;
In the obstacle determination step, when it is determined in the rain determination step that the rain state is present, the obstacle determination is performed based on the detected image that has been subjected to the rain removal process in the rain removal processing step. A featured obstacle detection method.   An obstacle detection program for causing a computer to execute each step according to claim 14.

Images